For measurements taken over a decade at the coastal Danish site
Høvsøre, we find the variance associated with wind speed events from
the offshore direction to exceed the prescribed extreme turbulence model
(ETM) of the International Electrotechnical Commission (IEC) 61400-1 Edition 3 standard for wind turbine safety. The
variance of wind velocity fluctuations manifested during these events is not
due to extreme turbulence; rather, it is primarily caused by ramp-like
increases in wind speed associated with larger-scale meteorological
processes. The measurements are both linearly detrended and high-pass
filtered in order to investigate how these events – and such commonly used
filtering – affect the estimated 50-year return period of turbulence levels.
High-pass filtering the measurements with a cutoff frequency of

The International Electrotechnical Commission (IEC) design standard for wind turbine safety

The IEC standard recommends the uniform-shear spectral turbulence model of

A further investigation of 10 min turbulence measurements exceeding the ETM
level is needed to identify what kind of flow causes these extreme events and
how they influence the estimated turbulence level at a given site.
Fluctuations associated with mesoscale meteorological motion can have periods
in the range of a minute up to hours

In this paper we aim to find and examine events for which the 10 min variance
exceeds the ETM level. However, here we consider them to be nonturbulent events,
as they are caused by a ramp-like increase in wind speed associated with
larger-scale meteorological processes, which may be observed offshore or high
above the surface layer. We use measurements from the measurement site
Høvsøre, focusing on the western (offshore) sectors. We demonstrate how
these events influence the estimate of 10 min turbulence levels with a
50-year return period. This is done for the raw, linearly detrended,
and high-pass-filtered measurements. The observed events are simulated by
incorporating measured time series using a constrained simulation approach
in order to get a realistic representation of the flow involved. The
generated wind field realizations are fed to an aeroelastic model

The data analysis and load simulations are based on measurements from the Høvsøre Test Centre for Large Wind Turbines in western Denmark. Located over flat terrain 1.7 km east of the coastline, the site offers low-turbulence, near-coastal wind conditions. The site consists of five wind turbines arranged in a single row along the north–south direction and multiple measurement masts.

The primary data source used in this paper is a light mast

The light mast has aircraft warning lights on the top.

placed between two of the wind turbines. This mast has cup anemometers and wind vanes at 60, 100, and 160 m heights installed on southward-pointing booms. The measurements span a 10-year period from November 2004 to December 2014, and the recording frequency is 10 Hz. The light-mast data are compared with data from the main Høvsøre meteorological mast, which is located south of all wind turbines and approximately 400 m south of the light mast, as may be seen in Fig.We consider measurements only from the western sector, with 10 min mean wind
direction between 225 and 315

For the selection of the extreme variance events the 10 min standard
deviation of the wind speed measurements is compared to the extreme
turbulence model in the IEC 61400-1 standard

The IEC standard has three turbulence categories: A, B, and C, with A being
the highest reference turbulence intensity and C the lowest. The
corresponding reference TI for each class may be seen in
Table

The dots correspond to the 10 min standard deviation of the wind speed
as a function of

The IEC turbulence classes and associated turbulence intensities.

Figure

Comparison of horizontal wind speed measurements at the
meteorological mast (green curve) and the light mast (blue curve). The
measurement height is 100 m at both masts, which are separated by

The data set used for the data analysis and simulation is composed of the 10 Hz measurements from cup anemometers and wind vanes on the light mast in Høvsøre.

The measurements shown earlier in Figs.

The high-pass filtering is performed with a second-order Butterworth filter

Fluctuations with a
period of 300 s at 4–25 m s

Here we use the inverse first-order reliability method (IFORM) to estimate
the 50-year return period contour corresponding to the joint description of
turbulence (

The first step in the IFORM analysis is to find the joint probability
distribution

Here we use
a three-parameter Weibull distribution. This is done because after filtering out
measurements with errors and missing periods, the lowest mean wind speed is
2.2 m s

In Fig.

The mean and standard deviation of

The next step in the IFORM analysis is to obtain a utility “reliability
index”

The 50-year return period contours based on the measurements (green
curves) and the IEC expressions (blue curves). The grey dots show the
measurements.

Figure

Note that some measurement points have been removed due
to measurement errors; therefore, the points are fewer than in
Fig.

The peak and the corresponding location of each event are identified in the
following way: a moving average is subtracted from the wind speed signal and
the maximum value of the differences identified:

Applying the selection criteria described in Sect.

Wind turbine response in the time domain is calculated with HAWC2

All the load simulations are performed using the DTU 10 MW reference wind
turbine (RWT), which is a virtual wind turbine model based on
state-of-the-art wind turbine design methodology. The main characteristics of
the RWT may be seen in Table

The main characteristics of the reference wind turbine.

The Mann spectral turbulence model

The Mann model is based on an isotropic von Kármán turbulence
spectral tensor, which is distorted by vertical shear caused by surface
friction. Assumptions of constant shear and neutral atmospheric conditions in
the rapid distortion limit are used to linearize the Navier–Stokes equations,
which may then be solved as simple linear differential equations. The
solution results in a spectral tensor that may be used in a Fourier
simulation to generate a random field with anisotropic turbulent flow. The
Mann model contains three parameters, as described below.

The IEC-recommended values of the parameters are

The DLC is simulated based on the setup described in

In contrast with

The aim here is to generate
turbulence simulations resembling the measured wind field of the extreme
variance events. This is done by constraining the synthesized turbulence
fields. The constraining procedure involves modifying the time series to
represent the most likely realization of a random Gaussian field that would
satisfy the constraints using an algorithm described in

the constrained field,

the source field,

the residual field, which is the difference between the constrained
field and the source field,

A realization of the constrained field is generated by adding the conditional
ensemble mean of the residual field to the source field:

Here the constraints consist of the

In Fig.

Comparison between

Comparison of unconstrained and constrained stream-wise (

Figure

For the purpose of load simulations, six different constrained turbulence seeds are generated from each extreme variance event time series. Although applying the constraints makes the turbulence boxes similar in general, there are differences in the parts of the boxes that are far from the constraint locations. As a result, there will be a seed-to-seed variation in loads simulated with constrained turbulence boxes, but they are much smaller than what is seen in the unconstrained case.

In this section we compare the design load levels of the two simulation sets: DLC 1.3 and the constrained simulations with the extreme variance. DLC 1.3 consists of 72 simulations (six seeds per 12 wind speed bins) and the constrained simulations consist of 264 simulations (six seeds per 44 extreme variance events).

In Fig.

The mean standard deviation of the

In Fig.

The mean extreme moments from IEC DLC 1.3 (grey dots) and the mean extreme loads from the constrained simulations (blue dots).

Figure

Figure

Figure

Figure

The extreme tower-top tilt, yaw, and tower-base side–side moments show a
general increase with wind speed. The extreme blade root flap and tower-base
fore–aft moments peak around rated wind speed. For the extreme blade root
edge moment it is seen that the loads peak around rated wind speed for both
simulation sets, but the main difference is that after 16 m s

The highest mean extreme moments for different load components

Table

In the following, examples of 10 min time series from DLC 1.3 and constrained simulation sets are shown side by side for comparison and a demonstration of the differences in the wind turbine response to different types of wind regime. A comparison is made for the tower-base fore–aft moment, wherein the characteristic extreme loads from the different simulation sets are of similar magnitude. We also consider and compare the tower-top tilt and yaw moments, which give the largest differences between the two simulation sets.

Comparison of a DLC 1.3 time series

First, we compare two time series giving some of the highest extreme tower-base fore–aft moments from each simulation set. For DLC 1.3 in
Fig.

Comparison of a DLC 1.3 time series

In Fig.

In the load time series comparison, the general differences in the wind
turbine response of the two simulation sets are visualized; for the
constrained simulations the peak loads are distinguishable and occur because
of the velocity increase associated with the ramp-like event. The
discrepancies between the two simulation sets for the extreme tower-top loads
indicate that the short-term wind field variability across the rotor is
generally higher in the stationary turbulence simulation than for the
constrained simulations. As shown in the time series comparison of
Fig.

The load simulation results show that the extreme turbulence case DLC1.3 indeed covers the load envelope caused by extreme variance events. However, the differences seen in the time series and in the load behavior indicate that extreme variance observations as events are entirely different from situations with stationary, homogeneous turbulence. This questions the basis for the definition of the IEC extreme turbulence model (ETM), which is defined in terms of the statistics of the 10 min standard deviation of wind speed. As most observations of the selected extreme variance events include a short-term ramp event, it would perhaps be more relevant to compare these events with other extreme design load cases in the IEC standard, e.g., the extreme coherent gust with direction change, extreme wind shear, or the extreme operating gust. Since these are the absolute highest variance events observed at Høvsøre during a 10-year period, they would also appear in the site-specific definition of the ETM. Therefore, it may be necessary to exclude or reassign such events to the relevant load case type. The design and cost of a wind turbine may depend on how this consideration is done.

In the current study we generate Gaussian turbulence fields only, though it
is known that atmospheric turbulence can exhibit some non-Gaussian character

It was seen in the IFORM analysis in Sect.

The main objective of this study is to investigate how extreme variance
events influence wind turbine response and how it compares with DLC 1.3 of
the IEC 61400-1 standard. The selected extreme events are measurements of the
10 min standard deviation of horizontal wind speed that exceed the values
prescribed by the ETM and include a sudden velocity jump (ramp event,
transients in the turbulent flow), which is the main cause of the high
observed variance. The events were simulated with constrained turbulence
simulations in which the measured time series were incorporated into turbulence
boxes for load simulations in order to make a realistic representation of the
events, including short-term ramps and coherent flow in the lateral
direction as was seen in the comparison of measurements between the two masts
in Fig.

Load calculations of the simulated extreme events were made in HAWC2 and
compared to load calculations with stationary homogeneous turbulence
according to DLC 1.3. To summarize, we have found the following.

The extreme variance events are large coherent structures, observed simultaneously at two different masts with a 400 m (lateral) separation.

Most extreme variance events include a sharp wind speed increase (short-time ramp), which is the main source of the large observed variance.

High-pass filtering with a cutoff frequency of

Compared with the DLC 1.3 of the IEC standard, the extreme loads are on average lower for the extreme variance events in the coastal and/or offshore climate and heights considered.

For 10 min mean wind speeds of 8–16 m s

Future related work includes further analysis and characterization of extreme variance events. In particular, ongoing work involves extreme short-term shear associated with such events and directional change. Load simulations of the events may be compared with other extreme DLCs from the IEC standard.

The high-frequency measurements used for the data processing in Sect. 3 are stored at DTU Wind Energy in a SQL database that is not publicly accessible. The HAWC2 simulation outputs and wind speed inputs (turbulence boxes) are available as binary files upon request to Ásta Hannesdóttir (astah@dtu.dk).

ÁH performed the data analysis and simulations. ÁH made all figures. MK provided guidance and comments. ND developed the code that is used to perform constrained turbulence simulations. ÁH prepared the paper with contributions from the coauthors. This work is part of ÁH's PhD under the supervision of MK.

The authors declare that no competing interests are present in this work.

The authors would like to thank Anand Natarajan and Jakob Mann for constructive comments and discussion. Ásta Hannesdóttir would also like to acknowledge Jenni Rinker and David Verelst for HAWC2 assistance.

This paper was edited by Joachim Peinke and reviewed by three anonymous referees.